[ADS- L] Ev idence for DECIMATE ( 'one in ten')

[Jesse Sheidlower]

On Mon, Jan 07, 2008 at 10:40:19PM -0500, Douglas G. Wilson wrote:
> >Yes, there's a difference between "_decimate_ has never been
> >found in the nonhistorical sense 'kill 10% of'" and "Using
> >full-text databases and scanning the entire body of English
> >writing, we can find three examples of _decimate_ in the
> >nonhistorical sense 'kill 10% of'".  But it's a distinction
> >without a difference. The frequency of this sense is
> >statistically zero.
>
> On what basis is this assessment made?

I've looked for it in the Oxford Corpus; I've done my own
searching in Google Books and other sources; and I've looked
at discussions of this in the Usenet newsgroup
alt.usage.english, where people tried hard to prove that this
sense exists.

> The few examples I presented were not scrounged with titanic effort
> from among thousands of citations, they were retrieved from Google
> Books in maybe 30 minutes by an amateur dilettante.
>
> I just now searched Google Books for "decimated" from 1850 to 1870,
> requiring limited preview at least (no snippets). About 800 items
> appeared. I don't have time to look at 800 so I went down the list
> and took the first 50 which had the sense "kill a portion" or so. I
> tried to exclude duplicates and I excluded incomprehensible and
> illegible results as well as those with other meanings (related,
> e.g., to tithing or weights and measures) and those where the word
> itself was under discussion.
>
> Here are the probable meanings according to my judgement:
>
> A. Kill a large arbitrary fraction
> B. Kill [about] 1/10
> C. A or B (I can't tell which)
> D. Kill 1/10 as military punishment
> E. Kill 9/10
>
> Numbers:
>
> A 23 (46%)
> B 10 (20%)
> C 8 (16%)
> D 6 (12%)
> E 3 (6%)
>
> Anybody else can perform a comparable experiment. Results will vary,
> of course. In particular the above A, B, C groups can change
> considerably in relative size depending on how hard one tries to
> force his favorite interpretation on each example.

I've just done this again, albeit haphazardly, with somewhat
different parameters, and must confess that I can't come close
to your numbers here.  Most of them seem to mean 'kill a lot
of', no more.

I value the evidence, and I'd have no qualms about changing my
views if the evidence doesn't support them. But I don't see
this from my searches.

To reply to Joel in a different message: If Murray had had a
real example of this, it would be in the entry or in the OED
Archives. It's in neither. It's possible that he was aware of
evidence that wasn't at hand, but it's more likely, as MWDEU
has it, that he assumed the evidence on etymological grounds.

Jesse Sheidlower
OED

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org